Efficient Monitoring of Atmospheric State

A dense, long-term and efficient monitoring of the atmospheric state is rarely met in real-life conditions except for experimental campaigns. Automatic weather stations can contribute, but insufficient spatial coverage or bad distribution of them are typical problems, let alone temporal gaps and discontinuities due to sensor failures or human aspects (poor maintenance). Given the absence of a dense in-situ network or weather radar scans over our area of study that could provide a necessary insight, we relied upon high resolution Numerical Weather Predictions (NWP) (2 km) from our convection-permitting operational configuration of WRF-ARW model [86 ]. The model is initialized daily with the latest available analysis and after the exclusion of the first few hours (spin-up time) in order to reach a statistical equilibrium, the following 24-hour estimates are utilized.
While this grid spacing may appear quite coarse when compared to the resolution of the EO-derived products, we should consider that this is an outcome of NWP simulations. We are able to provide estimates of atmospheric parameters every 2 km over regions that are heavily under-monitored (in-situ weather station can be available every 100 km over croplands). This scale is considered high resolution in NWP terms and the resolving of cloud microphysical processes, such as convection, starts to happen explicitly under this spatial threshold which is particularly important to resolve fine atmospheric processes on a local scale without having to rely on parameterization schemes. A 2 km forecast fulfils our needs given that the physiographic characteristics of the crop regions we focus upon are not areas of high topographical complexity, so great gradients are not expected.
The specific parameters that were used in our pipeline were an outcome of consultation with agronomist experts and systematic literature review upon their correlation with the evolution of cotton [54 (link)]. They include Air Temperature, Surface (skin) Temperature, 0–10 cm Soil Temperature and Moisture, Precipitation and Incoming Shortwave Radiation. Growing Degree Days (GDD) is additionally computed, as it is one of the most essential indicators of phenology. Inspecting the GDD equation in Table 4, T_max and T_min are the maximum and minimum daily air temperatures at 2m (from surface) and T_base is the crop’s base temperature (15.6°C). The latter is defined as the temperature below which cotton does not develop. The GDD variable is also known as thermal time and is an indicator of the effective growing days of the crop [87 (link)].